1. Field of the Invention
The present invention relates to a method and apparatus for measuring 3-D (Three-Dimension) information of an object from a pair of object images, which is obtained by capturing the object from two different directions.
2. Description of the Related Art
A xe2x80x9cstereoxe2x80x9d, or so called xe2x80x9cbinocular visionxe2x80x9d is known as a method for measuring 3-D information of an object, such as position and form, from images. According to the stereo method, the object is captured from two different directions so that a pair of images is obtained. Then, a pair of corresponding points, each of which is an image of one point of the 3-D object, is obtained from the pair of images. Based on the pair of corresponding points and the distance between two capturing points, the position of the object is calculated using triangulation. The stereo method is, for example, utilized for photogrammetry, such as aerial photogrammetry and photogrammetry for a traffic accident spot.
When calculating the position of the object, it is important to detect the pair of corresponding points correctly. This corresponding points determining problem has been an important technical problem in the stereo method and various ways to overcome it have been proposed.
Among some of the ways is a way that detects the pair of corresponding points on the basis of a specific portion of the object, such as an edge or a ridge-line. However, when the shape of the object is a curved surface, such as a cylinder, as the specific portion on the object cannot be easily identified, incorrect corresponding points are detected, or the corresponding points cannot be detected. Consequently, the actual 3-D information of the object cannot be obtained.
Therefore, an object of the present invention is to provide a method and apparatus for correctly detecting a pair of corresponding points and measuring 3-D information of the object when an object has a curved surface.
An apparatus for measuring three-dimension information of an object according to the present invention has an image reader, a corresponding point detector, and a three-dimension information calculator. This apparatus is applied for photogrammetry, or computer vision in the AI (Artificial Intelligence) field. For example, in the case of photogrammetry for a traffic accident spot, an exclusive still camera with an image sensor is used. The camera is arranged at two capturing points in order to capture the object from two directions. The object to be captured is cylindrical, in other words, the object is a body of revolution. The object has substantially rotational symmetry with respect to a central axis of the object, and a given cross-section perpendicular to the central axis is a circle. For example, the object is a cylinder-shaped object or a frustum-shaped object and so on. When the object is captured, the image reader reads a pair of images recorded in a recording medium. For example, in the case of photogrammetry, the pair of images is recorded in a memory card detachably installed in the camera and apparatus. When the memory card is installed in the apparatus, the pair of images is read from the recording medium and is then temporarily stored in a memory, such as a RAM (Random Access Memory), by the image reader. The pair of images is obtained by capturing the object from two capturing points such that apair of contour lines, called xe2x80x9coccluding contoursxe2x80x9d is reflected in each pair of images.
The object image in each of the obtained pair of images has line symmetry with respect to an imaginary projected image of the central axis, which bisects the object image. According to the apparatus, the corresponding point detector detects at least one pair of corresponding points on a pair of bisecting lines defined in the pair of images. The pair of corresponding points is an imaginary pair of images of a measuring point positioned on the central axis, and the pair of corresponding points is uniquely determined in the pair of images. Each of the pair of bisecting lines is an imaginary projected image of the central axis. The three-dimension information calculator calculates a three-dimension position of the at least one measuring point on the basis of the at least one pair of corresponding points. To calculate the three-dimension information, a triangulation method is applied.
According to the present invention, the imaginary pair of bisecting lines is defined, and then at least one pair of corresponding points is detected on the pair of bisecting lines. Consequently, the pair of corresponding points is detected correctly and 3-D information of the object is obtained correctly.
To detect the at least one pair of corresponding points correctly, preferably, an epipolar line, which is used in the photogrammetry or in the computer vision, is defined. The corresponding point detector defines at least one first image point on one of the pair of bisecting lines in one of the pair of images, and sets at least one epipolar line, corresponding to the at least one first image point, in the other image of the pair of images. Then, the corresponding point detector defines at least one second image point, which is a crossing point of the other of the pair of bisecting lines and the at least one epipolar line. The at least one first image point and the at least one second image point is defined as the at least one pair of corresponding points.
In the case of the photogrammetry, usually, an operator performs a given process for calculating the 3-D information using peripheral equipment, such as a keyboard or mouse. To calculating the 3-D information with the support of an operator, preferably, the apparatus includes a display and an indicating point inputting device. When the pair of images is read from the recording medium, the pair of images is displayed on the display. The indicating point inputting device is operated for inputting two pairs of indicating points on the pair of occluding contours in each of the pair of images. The corresponding point detector detects the two pairs of indicating points input by the operator, and calculates the pair of bisecting lines and the at least one pair of corresponding points in accordance with the two pairs of indicating points in each of the pair of images. In this case, the at least one pair of corresponding points is not automatically detected by a given image-processing method, such as an edge detecting process, but detected on the basis of the input points.
To measure a radius of the object with the 3-D position, preferably, the apparatus includes a radius calculator. In this case, the corresponding point detector detects two pairs of corresponding points, and the three-dimension information calculator calculates the positions of two measuring points on the basis of the two pair of corresponding points. The radius calculator firstly calculates a plane, in which a vector, perpendicular to the central axis and passing one of the two capturing points, is a normal vector and the central axis is included, from the positions of the two measuring points. Next, the radius calculator calculates an edge point, which is on the plane and a curved surface of the object, on the basis of an image point. The position along the central axis with respect to the edge point is the same as the position of one of the two measuring points. The image point is on one of the pair of occluding contours and corresponds to the edge point. Then, the radius calculator calculates a radius from the edge point and one of the two measuring points.
According to another aspect of the present invention, a method for measuring three-dimension information of an object includes steps of: 1) reading a pair of images recorded in a recording medium, the pair of images being obtained by capturing the object from two capturing points, the object being cylindrical, a pair of occluding contours being reflected in each of the pair of images; 2) detecting at least one pair of corresponding points, which is an imaginary pair of images of at least one measuring point positioned on the central axis and is uniquely determined in the pair of images, on a pair of bisecting lines defined in the pair of images, each pair of bisecting lines being imaginary projected image of the central axis; 3) calculating a three-dimension position of the at least one measuring point on the basis of the at least one pair of corresponding points by applying a triangulation method.
According to another aspect of the present invention, a memory medium that stores a program for measuring three-dimension information of an object. The program includes steps of: 1) reading a pair of images recorded in a recording medium in the apparatus, the pair of images being obtained by capturing the object from two capturing points, the object being cylindrical, a pair of occluding contours being reflected in each of the pair of images; 2) detecting at least one pair of corresponding points, which is an imaginary pair of images of at least one measuring point positioned on the central axis and is uniquely determined in the pair of images, on a pair of bisecting lines defined in the pair of images, each of the pair of bisecting lines being imaginary projected image of the central axis; 3) calculating a three-dimension position of the at least one measuring point on the basis of the at least one pair of corresponding points by applying a triangulation method.